Starter
$0
- Bring your own model / API key
- Public artifacts
Pricing
Open source.
Free for everyone.
Run it anywhere — your own computer, your VPC, or our cloud. Bring your own LLM API keys, or use our cloud (5M tokens included).
- Run anywhere
- 5M tokens included
Plans
Start free. Go Pro for $9.95 a month.
Bring your own model and API key on Starter, or go Pro for the full Model Router across every frontier and open model — 5M tokens included, billed monthly, cancel anytime.
Pro
$9.95/ month
- Everything in Starter, plus
- Access to all frontier models
- 5M tokens included
- Private artifacts
Model router
One endpoint. Every model worth using.
The MindsHub router gives your agent a single connection to frontier
and open-source models. Reference models by latest:*
aliases so your agent stays current as providers ship new versions.
Unified access
One API surface for Anthropic, OpenAI, Google, and popular open models — no per-provider keys to manage.
Latest aliases
Point at latest:sonnet, latest:gpt,
latest:gemini, and the router resolves to the
current version.
Mix and match
Use strong reasoning models for planning and cheaper models for routine steps — all from the same endpoint.
Available models
Frontier and open-source, ready to call.
Every model worth using, through the MindsHub Model Router — reference
them by latest:* aliases that stay current as providers
ship new versions.
Anthropic Frontier
| Alias | Resolves to |
|---|---|
latest:sonnet | claude-sonnet-4-6 |
latest:opus | claude-opus-4-8 |
latest:haiku | claude-haiku-4-5-20251001 |
OpenAI Frontier
| Alias | Resolves to |
|---|---|
latest:gpt | gpt-5.5 Low reasoning effort — same as latest:gpt-low |
latest:gpt-low latest:gpt-medium latest:gpt-high | gpt-5.5 With matching reasoning effort level |
latest:gpt-codex | gpt-5.3-codex |
latest:gpt-mini | gpt-5.4-mini |
latest:gpt-nano | gpt-5.4-nano |
Google Frontier
| Alias | Resolves to |
|---|---|
latest:gemini | gemini-3.1-pro-preview |
latest:gemini-flash | gemini-3.5-flash |
Open source Open weights
| Alias | Resolves to |
|---|---|
latest:kimi | kimi-k2p6 |
latest:deepseek | deepseek-v4-pro |
latest:qwen | qwen3p6-plus |
FAQ
Common questions.
What is MindsHub?
MindsHub is the platform for open-source AI agents. Delegate whole tasks in Cowork, the unified workspace, or run your own agent with its native UI and harness.
How much does MindsHub cost?
It is $9.95/month with 5M tokens included — whether you use Cowork, the unified workspace, or run your own open-source agent ($9.95/month per agent). Billed monthly, cancel anytime. This is early-access pricing while we build usage-based billing.
What do the 5M tokens cover in practice?
Think of it as a generous monthly AI allowance. In everyday use that's plenty of delegated work — for example a weekly competitive digest, a handful of dashboards, a stack of document reviews, and recurring briefings. Heavier or premium-model tasks draw it down faster, and your usage is visible in your account.
Do I need API keys to start?
No. The Model Router is wired in, so there are no provider keys to set up. You can connect your own LLM accounts later if you want direct provider control.
What happens when usage-based billing ships?
Detailed, usage-based billing is coming soon. We'll publish a transparent rate card before anything changes, and early subscribers keep their intro rate.
Can I switch between Cowork and running my own agent?
Yes. One MindsHub account works for both — switch anytime.
Which open-source agents can I run?
Anton and Hermes run in Cowork today. OpenClaw, OpenClaw + GBrain, NanoClaw, and Hermes run standalone with their native harness — more on the way.
How do the latest:* aliases work?
Instead of pinning a specific model version, reference an alias like latest:sonnet or latest:gpt. The Model Router resolves it to the current pinned version of that family, so you pick up new releases without changes.